>I am objecting to embeddeding metadata in data. .... >I think we were just looking a different aspects of the elephant ;-)
I think you are right on both counts. Given current filesystems, I like the #! method. I tend to like approaches that have very low entrance access fees and can scale up. Kinda like python's "hello world" versus Java's. It seems as though our managing of complexity fails exponentially the more cruft you add to things. The more you can keep things simple, the further you can avoid the inflection point. john -- http://mail.python.org/mailman/listinfo/python-list
